NASP-T: A Fuzzy Neuro-Symbolic Transformer for Logic-Constrained Aviation Safety Report Classification

2025-10-06

AI Summary

Deep Transformer models for multi-label classification of aviation safety reports often violate domain-specific logical rules, compromising reliability and interpretability. Method: This paper proposes a neuro-symbolic integration framework that embeds Answer Set Programming (ASP) into the Transformer training pipeline. Domain knowledge is formalized in ASP, rule consistency is verified via the Clingo solver, and a differentiable fuzzy-logic regularization term—combined with rule-guided data augmentation—is introduced to enforce logical constraints during end-to-end learning. Contribution/Results: The approach preserves full end-to-end trainability while significantly enhancing model reliability and interpretability. Experiments on aviation safety report data demonstrate consistent improvements over strong baselines: micro-F1 and macro-F1 both increase, and the logical rule violation rate drops by 86%, achieving superior trade-offs between classification accuracy and domain logic compliance.

Technology Category

• Machine Learning: Neuro-Symbolic Learning
• Natural Language Processing: Safety and Robustness
• Constraint Satisfaction and Optimization: Satisfiability Modulo Theories

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Security and Privacy: Data transparency and provenance

Abstract

Deep transformer models excel at multi-label text classification but often violate domain logic that experts consider essential, an issue of particular concern in safety-critical applications. We propose a hybrid neuro-symbolic framework that integrates Answer Set Programming (ASP) with transformer-based learning on the Aviation Safety Reporting System (ASRS) corpus. Domain knowledge is formalized as weighted ASP rules and validated using the Clingo solver. These rules are incorporated in two complementary ways: (i) as rule-based data augmentation, generating logically consistent synthetic samples that improve label diversity and coverage; and (ii) as a fuzzy-logic regularizer, enforcing rule satisfaction in a differentiable form during fine-tuning. This design preserves the interpretability of symbolic reasoning while leveraging the scalability of deep neural architectures. We further tune per-class thresholds and report both standard classification metrics and logic-consistency rates. Compared to a strong Binary Cross-Entropy (BCE) baseline, our approach improves micro- and macro-F1 scores and achieves up to an 86% reduction in rule violations on the ASRS test set. To the best of our knowledge, this constitutes the first large-scale neuro-symbolic application to ASRS reports that unifies ASP-based reasoning, rule-driven augmentation, and differentiable transformer training for trustworthy, safety-critical NLP.

Problem

• Integrating domain logic constraints into transformer models for classification
• Reducing rule violations in safety-critical aviation report analysis
• Combining symbolic reasoning with neural networks for trustworthy NLP

Innovation

• Hybrid neuro-symbolic framework integrating ASP with transformers
• Rule-based data augmentation generating logically consistent samples
• Fuzzy-logic regularizer enforcing rule satisfaction during fine-tuning